Exposure ManagementPlatform

Users were used to looking at security problems one list at a time. The new platform showed how everything connects. The design job was moving people across that mental-model shift without losing their trust.

Result

21 → 6

clicks on the most time-consuming routes

  • RoleDesign Lead
  • Timeline8 weeks
  • WhenDec 2024 – Jan 2025
  • Team6 disciplines
  • UsersSecurity analysts, security teams, CISOs
  • PlatformWeb app · enterprise
Exposure/Overview/Finance team
Live · scanned 12 min ago

Exposure score

72/100

▼ 6 in 3 weeks · High

Internet-facing assets34%

Critical weaknesses28%

Failing controls22%

Account hygiene16%

Cloud control coverage

84%

▲ 3.2% this week

Attack surface

1,247

42 new assets

Open findings

318

12 since yesterday

Top findings

FindingSeverityStatus
CriticalOpen
CriticalOpen
HighOpen
HighIn progress
MediumResolved
LowOpen

Exposure path

5 steps · 1 failing

Remote code execution · api.gateway

  1. SourceInternet
  2. Assetcdn-edge-02
  3. Assetapi.gateway
  4. WeaknessCVE-2024-23113
  5. ControlPCI-DSS 6.3.3✕ failing
  6. TeamFinance

Working recreation. Press L for the decision behind each part.Open “Design notes” in the footer for the decision behind each part.

NDA Interactive recreation with sample data. These are not production screens or customer data.

In 30 seconds

  1. 1

    The problem

    Customers of two separate security tools had to move from reading lists to reading relationships, without losing trust in a platform they relied on.

  2. 2

    What I did

    I sequenced the shift: familiar entry points kept, shorter routes, backtracking through every relationship view, scores that show their working, and severity that never relies on colour alone.

  3. 3

    The outcome

    The most time-consuming routes went from 21 clicks to 6, and people no longer lose their way six or seven clicks deep.

Jump toKey decisions Results

01The problem

Three teams needed three different things from the same release

Prevalent's cloud security and attack surface products showed risk as separate lists. A shared data model could connect them, linking a security issue to the device, policy and team it affects, and become the foundation for everything Prevalent built next, including Navigator and Studio.

I owned the platform UX strategy, from the v2.0 navigation and product structure to the customisable overview dashboard, exposure paths, score derivation and an accessible severity system.

Three needs competed: existing customers needed continuity, sales needed a visible platform story, and engineering was still building how the data came together. We agreed up front that adoption risk would decide ties. That turned recurring debates into quick decisions.

Constraints

Existing customers
No forced migration of the workflows they relied on.
Sales
Needed visible, differentiated capability to position the platform as exposure management.
Data still being built
Engineering limits on how quickly and completely relationships could be surfaced.
Accessibility
Dense, colour-coded security data had to remain legible without colour.

Customers had to learn a new way of reading risk

Findings were judged without asset or business context, and scores couldn't be traced back to the data behind them. Connecting everything solved that on paper. But relationships are harder to read than lists.

Two risks dominated: overwhelming users with interconnected data before they were ready for it, and losing their trust if exposure calculations couldn't be explained.

The briefAdd a map view to the product.

The brief, rewrittenMove users from lists to relationships without losing them on the way.

02Research

Analysts had the data. They were missing the context

The core insight came early. Research then focused on how much context helps before it starts to hurt.

How

Workflow mapping
Mapped how existing users sort, investigate and report on issues, including the workarounds between the two products.
Stakeholder sessions
Working sessions with sales, engineering and customer success to surface the competing needs early.
Terminology testing
Tested new concepts (exposure, path, blast radius) for whether users could explain them back in their own words.
Depth testing
Followed people through relationship views at increasing depth, to find where they lost their way.

What we learned

  1. 01
    Context, not data

    Users already had more findings than they could act on. What they lacked was why a finding mattered here.

    Which led to Every finding resolves to its asset, control and business unit.

  2. 02
    Lost after six or seven clicks

    Following relationships from finding to asset to control, people lost track of where they were and how they got there after six or seven clicks.

    Which led to Backtracking everywhere, including inside complex relationship charts, so the path you came by is always visible.

  3. 03
    The slowest routes took 21 clicks

    Counting clicks on the most time-consuming routes through the product gave a concrete baseline to design against.

    Which led to The v2.0 navigation, rebuilt around those routes.

  4. 04
    "Why is this critical?" comes first

    The most common question about any score was how it was derived.

    Which led to Scores that decompose into contributing factors and trend.

  5. 05
    Familiarity is trust

    Users were willing to adopt new views when they started from screens they already knew.

    Which led to The two products kept their place in the navigation, with connected context added inside.

03Approach

Principles for this product

Anchor the new in the familiar
New concepts attach to existing navigation, language and workflows.
Show the working
Any number on screen can be traced to the data behind it.
Sequence, don't flood
Introduce new ideas in an order people can absorb.

I tested three ways to show relationships, at increasing depth

I explored map, table and path representations of the same exposure, and tested each at increasing depth. The validation rule was simple: a design only advanced if users could explain exposure back in their own words. Preference didn't count.

Tried and set aside

Relationship map first
Impressive in demos, but comprehension fell as depth increased.
New unified navigation
Built around the map, with a relearning cost for every existing customer.
Colour-only severity
Fails WCAG 1.4.1, colour-blind analysts, dim control-room screens and black-and-white exports.

Sketches

Redrawn for this portfolio. The original work is under NDA.

  • Hand-drawn sketch of the Exposure Management overview, with summary numbers, exposure by environment, top risk categories, a trend chart and a Top exposures table that opens an asset details drawer.
    The overview: find, prioritise and reduce risk. Every row in Top exposures should show business context, why it matters and a recommended action.
  • Sketch comparing a 21-step route with a 6-step route.
    Counting clicks on the slowest routes gave a baseline to design against (21 before, 6 after the v2.0 navigation).
  • Sketch of an asset details drawer open beside the Top exposures list.
    Details open in a drawer, so the list stays in view and nobody loses their place.
  • Sketch comparing colour-only severity dots with shape-and-label severity.
    Severity carried by shape and word, with colour only as backup.

04Decisions

  1. Decision 01Progressive disclosure of relationshipsWhat I chose A linear exposure path for the selected finding, with the full map available but secondary.
    What I turned down
    An exploratory relationship map as the primary view.
    Why
    In testing, comprehension fell as map depth increased. The path answers "how is this exposed?" directly; the map remains for experts who ask for it.
    What it cost
    The platform's most impressive view became its least prominent one.
  2. Decision 02A way back from anywhereWhat I chose Backtracking on every relationship view, including complex relationship charts, showing the path the person took to get there.
    What I turned down
    Relying on the browser back button and breadcrumbs that only reflect the page hierarchy.
    Why
    People lost their way six or seven clicks into the map. Exploring relationships is only safe if getting back is effortless.
    What it cost
    Another layer of chrome on views that are already dense.
  3. Decision 03Preserve familiar entry pointsWhat I chose The two existing products stay in the navigation where users expect them, with connected context added inside each.
    What I turned down
    A new unified navigation built around the map.
    Why
    Relearning cost would have stalled adoption among existing customers, the people the platform most needed to keep.
    What it cost
    Slowed how quickly sales could show the full platform story.
  4. Decision 04Severity without relying on colourWhat I chose Each severity has a distinct shape and a text label, with colour as reinforcement.
    What I turned down
    Colour-only badges and dots.
    Why
    Colour-only signals fail WCAG 1.4.1, fail colour-blind analysts, and degrade on dim control-room screens and black-and-white exports.
    What it cost
    Slightly denser rows in tables built for scanning.
  5. Decision 05Scores that show their workingWhat I chose The exposure score shows its trend and decomposes into the factors behind it.
    What I turned down
    A single headline number with a tooltip definition.
    Why
    "Why is this critical?" was the most common question. A score that can't be explained can't be defended upward.
    What it cost
    More on screen around every score. Worth it, because the score drives prioritisation.

A score people act on has to show its working

Exposure Management is the foundation Navigator and Studio are built on rather than an AI feature itself. But it is where the explainability patterns started: a computed score that ranks risk is only useful if the person ranking work can see why.

  1. Mental models
    From lists to relationships, in sequence

    Posture, then assets, then exposure, each introduced from an entry point users already knew, so the new model extended the old one instead of replacing it.

  2. Explainability & trust
    Every number decomposes

    The exposure score carries its trend and contributing factors, answering "why is this critical?" without leaving the view. Any number on screen can be traced to the data behind it.

Grouped by the themes of Google’s People + AI Guidebook.

Deliberately left out

Full map exploration upfront
Kept for later phases, once users were fluent in relationship views.
Advanced configuration early
Deferred to avoid front-loading complexity on top of new concepts.
All relationships by default
Only relationships that support the current task are shown.

Sequenced rollout

  1. Phase 1

    Posture

    Cloud controls stay where users know them. Each policy now links to what it protects.

    Entry point Cloud controls

  2. Phase 2

    Assets & attack surface

    Attack surface joins the same view. Each device shows the policies and issues it connects to.

    Entry point Attack surface

  3. Phase 3

    Connected exposure

    Exposure paths show how a risk reaches the business. The full map stays optional.

    Entry point Exposure paths

Connected views were introduced in phases, each anchored to a familiar entry point.

05Testing

Depth was fine; getting lost wasn't

Relationships are the point of the platform, so the fix couldn't be to hide them. It had to be making every step reversible, and making the common routes short.

  1. Navigating relationships

    1. First version

      Drill from finding to asset to control to business unit, one view at a time.

    2. What testing showed

      After six or seven clicks people lost track of where they were and how they got there.

    3. What shipped

      Backtracking on every view, including complex relationship charts, so the path you came by is always visible and one step back is always available.

    First version: Drill from finding to asset to control to business unit, one view at a time.What shipped: Backtracking on every view, including complex relationship charts, so the path you came by is always visible and one step back is always available.
    Drag to compare. Sketches redrawn for this portfolio; the original work is under NDA.

06Results

One shared model, and nobody forced to relearn

  • 21 → 6Clicks

    On the most time-consuming routes, after the v2.0 navigation and multi-module structure.

  • 6–7 clicksWhere people got lost, now fixed

    The depth at which people lost their way. Backtracking now works everywhere, including inside relationship charts.

  • 1Shared model

    Two products brought together on one shared model, with every issue linked to the device, policy and team it affects.

  • 0Forced migrations

    Existing workflows preserved; connected views adopted from familiar entry points.

What I’d measure next

Time to first action
From finding to first action, to quantify what connected context saves against a baseline.

A new way of reading risk has to arrive in steps

Sequencing beats any single screen
Introducing new ideas in the right order did more for understanding than any one screen.
Test comprehension, not preference
The full map demoed brilliantly and performed poorly in tasks. Asking users to explain exposure back kept the design honest.
Set a baseline first
I'd measure how long it takes to act on an issue before launch, so the value of the redesign can be compared.
Next: open exploration for experts
Now that users read paths fluently, introduce the relationship map as an opt-in expert mode.
Next: what-if remediation
Simulate how patching one asset changes the score and the paths through it, to prioritise by impact.